added an optional third parameter to str_word_count(), which is used as

a list of characters which are considerd "is_alpha()". Thus V14GR4 can
be extracted as one word if the user supplies "14" as charlist.
FR #31560
# Nuno, are you wishing to document this one too? :)

diff --git a/ext/standard/string.c b/ext/standard/string.c
index 4052cf0b13fe72b98ed54415f89776dd5343d9f9..7d0df42e04d1db24f8e5eda8e9991e43ae741989 100644 (file)
--- a/ext/standard/string.c
+++ b/ext/standard/string.c
@@ -4643,7 +4643,7 @@ PHP_FUNCTION(str_shuffle)
 }
 /* }}} */
 
-/* {{{ proto mixed str_word_count(string str, [int format])
+/* {{{ proto mixed str_word_count(string str, [int format [, string charlist]])
        Counts the number of words inside a string. If format of 1 is specified,
        then the function will return an array containing all the words
        found inside the string. If format of 2 is specified, then the function
@@ -4656,38 +4656,44 @@ PHP_FUNCTION(str_shuffle)
 */
 PHP_FUNCTION(str_word_count)
 {
-       zval **str, **o_format;
+       zval **str, **o_format, **also_alphas;
        char *s, *e, *p, *buf;
+       char *p_also = NULL;
        int word_count = 0;
        int type = 0;
        int n_args = ZEND_NUM_ARGS();
 
-       if (n_args > 2 || n_args < 1 || zend_get_parameters_ex(n_args, &str, &o_format) == FAILURE) {
+       if (n_args > 3 || n_args < 1 || zend_get_parameters_ex(n_args, &str, &o_format, &also_alphas) == FAILURE) {
                WRONG_PARAM_COUNT;
        }
-       
-       if (n_args == 2) {
-               convert_to_long_ex(o_format);
-               type = Z_LVAL_PP(o_format);
+       switch (n_args) {
+               case 3:
+                       convert_to_string_ex(also_alphas);
+                       if (Z_STRLEN_PP(also_alphas)) {
+                               p_also = Z_STRVAL_PP(also_alphas);
+                       }
+               case 2:
+                       convert_to_long_ex(o_format);
+                       type = Z_LVAL_PP(o_format);
                
-               if (type != 1 && type != 2) {
-                       php_error_docref(NULL TSRMLS_CC, E_WARNING, "The specified format parameter, '%d' is invalid.", type);
-                       RETURN_FALSE;
-               }
+                       if (type != 1 && type !=2 && !(n_args >= 3 && type == 0)) {
+                               php_error_docref(NULL TSRMLS_CC, E_WARNING, "The specified format parameter, '%d' is invalid.", type);
+                               RETURN_FALSE;
+                       } 
+                       if (type) {
+                               array_init(return_value);
+                       }
        }
 
        convert_to_string_ex(str);
-       
        e = Z_STRLEN_PP(str) + (p = s = Z_STRVAL_PP(str));
-               
-       if (type) {
-               array_init(return_value);
-       }
-       
        while (p < e) {
                if (isalpha(*p++)) {
-                       s = p - 1;
-                       while (isalpha(*p) || *p == '\'' || (*p == '-' && isalpha(*(p+1)))) {
+                       s = p - 1; /* the word starts at s */
+                       while (isalpha(*p) || *p == '\'' || (*p == '-' && isalpha(*(p+1)))
+                                       || (p_also && php_memnstr(p_also, p, 1, p_also + Z_STRLEN_PP(also_alphas)))
+                                       )
+                       {
                                p++;
                        }
                        

diff --git a/ext/standard/tests/strings/str_word_count.phpt b/ext/standard/tests/strings/str_word_count.phpt
index 4c9c0780f63d9b0a2042d6e2637d2bd7071e0004..aa5eb9e87c1dcb26d9f069a0a063412607cca9b8 100644 (file)
--- a/ext/standard/tests/strings/str_word_count.phpt
+++ b/ext/standard/tests/strings/str_word_count.phpt
@@ -17,6 +17,26 @@ var_dump(str_word_count($str, array()));
 var_dump(str_word_count($str, $b));
 var_dump($str);
 
+$str2 = "F0o B4r 1s bar foo";
+var_dump(str_word_count($str2, NULL, "04"));
+var_dump(str_word_count($str2, NULL, "01"));
+var_dump(str_word_count($str2, NULL, "014"));
+var_dump(str_word_count($str2, NULL, array()));
+var_dump(str_word_count($str2, NULL, new stdClass));
+var_dump(str_word_count($str2, NULL, ""));
+var_dump(str_word_count($str2, 1, "04"));
+var_dump(str_word_count($str2, 1, "01"));
+var_dump(str_word_count($str2, 1, "014"));
+var_dump(str_word_count($str2, 1, array()));
+var_dump(str_word_count($str2, 1, new stdClass));
+var_dump(str_word_count($str2, 1, ""));
+var_dump(str_word_count($str2, 2, "04"));
+var_dump(str_word_count($str2, 2, "01"));
+var_dump(str_word_count($str2, 2, "014"));
+var_dump(str_word_count($str2, 2, array()));
+var_dump(str_word_count($str2, 2, new stdClass));
+var_dump(str_word_count($str2, 2, ""));
+
 ?>
 --EXPECTF--
 array(6) {
@@ -67,4 +87,194 @@ bool(false)
 Warning: str_word_count(): The specified format parameter, '0' is invalid. in %s on line 14
 bool(false)
 string(55) "Hello friend, you're  
-    looking          good today!"
\ No newline at end of file
+    looking          good today!"
+int(5)
+int(6)
+int(5)
+
+Notice: Array to string conversion in %s on line 21
+int(7)
+
+Notice: Object of class stdClass to string conversion in %s on line 22
+int(7)
+int(7)
+array(5) {
+  [0]=>
+  string(3) "F0o"
+  [1]=>
+  string(3) "B4r"
+  [2]=>
+  string(1) "s"
+  [3]=>
+  string(3) "bar"
+  [4]=>
+  string(3) "foo"
+}
+array(6) {
+  [0]=>
+  string(3) "F0o"
+  [1]=>
+  string(1) "B"
+  [2]=>
+  string(1) "r"
+  [3]=>
+  string(1) "s"
+  [4]=>
+  string(3) "bar"
+  [5]=>
+  string(3) "foo"
+}
+array(5) {
+  [0]=>
+  string(3) "F0o"
+  [1]=>
+  string(3) "B4r"
+  [2]=>
+  string(1) "s"
+  [3]=>
+  string(3) "bar"
+  [4]=>
+  string(3) "foo"
+}
+
+Notice: Array to string conversion in %s on line 27
+array(7) {
+  [0]=>
+  string(1) "F"
+  [1]=>
+  string(1) "o"
+  [2]=>
+  string(1) "B"
+  [3]=>
+  string(1) "r"
+  [4]=>
+  string(1) "s"
+  [5]=>
+  string(3) "bar"
+  [6]=>
+  string(3) "foo"
+}
+
+Notice: Object of class stdClass to string conversion in %s on line 28
+array(7) {
+  [0]=>
+  string(1) "F"
+  [1]=>
+  string(1) "o"
+  [2]=>
+  string(1) "B"
+  [3]=>
+  string(1) "r"
+  [4]=>
+  string(1) "s"
+  [5]=>
+  string(3) "bar"
+  [6]=>
+  string(3) "foo"
+}
+array(7) {
+  [0]=>
+  string(1) "F"
+  [1]=>
+  string(1) "o"
+  [2]=>
+  string(1) "B"
+  [3]=>
+  string(1) "r"
+  [4]=>
+  string(1) "s"
+  [5]=>
+  string(3) "bar"
+  [6]=>
+  string(3) "foo"
+}
+array(5) {
+  [0]=>
+  string(3) "F0o"
+  [4]=>
+  string(3) "B4r"
+  [9]=>
+  string(1) "s"
+  [11]=>
+  string(3) "bar"
+  [15]=>
+  string(3) "foo"
+}
+array(6) {
+  [0]=>
+  string(3) "F0o"
+  [4]=>
+  string(1) "B"
+  [6]=>
+  string(1) "r"
+  [9]=>
+  string(1) "s"
+  [11]=>
+  string(3) "bar"
+  [15]=>
+  string(3) "foo"
+}
+array(5) {
+  [0]=>
+  string(3) "F0o"
+  [4]=>
+  string(3) "B4r"
+  [9]=>
+  string(1) "s"
+  [11]=>
+  string(3) "bar"
+  [15]=>
+  string(3) "foo"
+}
+
+Notice: Array to string conversion in %s on line 33
+array(7) {
+  [0]=>
+  string(1) "F"
+  [2]=>
+  string(1) "o"
+  [4]=>
+  string(1) "B"
+  [6]=>
+  string(1) "r"
+  [9]=>
+  string(1) "s"
+  [11]=>
+  string(3) "bar"
+  [15]=>
+  string(3) "foo"
+}
+
+Notice: Object of class stdClass to string conversion in %s on line 34
+array(7) {
+  [0]=>
+  string(1) "F"
+  [2]=>
+  string(1) "o"
+  [4]=>
+  string(1) "B"
+  [6]=>
+  string(1) "r"
+  [9]=>
+  string(1) "s"
+  [11]=>
+  string(3) "bar"
+  [15]=>
+  string(3) "foo"
+}
+array(7) {
+  [0]=>
+  string(1) "F"
+  [2]=>
+  string(1) "o"
+  [4]=>
+  string(1) "B"
+  [6]=>
+  string(1) "r"
+  [9]=>
+  string(1) "s"
+  [11]=>
+  string(3) "bar"
+  [15]=>
+  string(3) "foo"
+}
\ No newline at end of file